检索结果-内蒙古大学图书馆

Towards optimized tensor code generation for deep learning on sunway many-core processor

Frontiers of Computer Science 2024年第2期18卷 1-15页

作者： Mingzhen LI Changxi LIU Jianjin LIAO Xuegui ZHENG Hailong YANG Rujun SUN Jun XU Lin GAN Guangwen YANG Zhongzhi LUAN Depei QIAN State Key Laboratory of Software Development Environment Beijing 100191China School of Computer Science and Engineering Beihang UniversityBeijing 100191China National University of Singapore Singapore 119077Singapore State Key Laboratory of Mathematical Engineering and Advanced Computing Wuxi 214000China Science and Technology on Special System Simulation Laboratory Beijing Simulation Center Beijing 100854China Department of Computer Science and Technology Tsinghua UniversityBeijing 100084China

The flourish of deep learning frameworks and hardware platforms has been demanding an efficient compiler that can shield the diversity in both software and hardware in order to provide application *** the existing deep learning compilers,TVM is well known for its efficiency in code generation and optimization across diverse hardware *** the meanwhile,the Sunway many-core processor renders itself as a competitive candidate for its attractive computational power in both scientific computing and deep learning *** paper combines the trends in these two ***,we propose swTVM that extends the original TVM to support ahead-of-time compilation for architecture requiring cross-compilation such as *** addition,we leverage the architecture features during the compilation such as core group for massive parallelism,DMA for high bandwidth memory transfer and local device memory for data locality,in order to generate efficient codes for deep learning workloads on *** experiment results show that the codes generated by swTVM achieve 1.79x improvement of inference latency on average compared to the state-of-the-art deep learning framework on Sunway,across eight representative *** work is the first attempt from the compiler perspective to bridge the gap of deep learning and Sunway processor particularly with productivity and efficiency in *** believe this work will encourage more people to embrace the power of deep learning and Sunwaymany-coreprocessor.

关键词： sunway processor deep learning compiler code generation performance optimization

来源：评论

学校读者我要写书评

暂无评论

A Review of Few-Shot Network Traffic Classification Based on Deep Learning

A Review of Few-Shot Network Traffic Classification Based on...

引用

2023 International Conference on Mechatronics and Intelligent Control, ICMIC 2023

作者： Yang, Kai Wang, Ruo Nan Meng, Qi Meng, Long Qi, Zan Li, Xue Fei, Jin Long State Key Laboratory of Mathematical Engineering And Advanced Computing PLA Strategic Support Force Information Engineering University Zhengzhou China

ISBN: (纸本)9781510668393

At present, deep learning has achieved excellent performance in the field of network traffic classification. However, deep learning relies on massive data-driven classification models. When the data set is small, it is usually hindered. In order to solve this problem, few-shot network traffic classification technology based on deep learning has been gradually studied. In this paper, we conducted a comprehensive survey to fully understand the few-shot traffic classification techniques. We first define the few-shot traffic classification problem. Based on how to solve the contradiction between the few-shot data set and the large number of parameters to be trained in the model, we classify the current few-shot traffic classification method based on deep learning from two perspectives: (1) Data augmentation method, which uses the method of expanding the data set to enhance the supervision experience. In this paper, the methods based on data enhancement are divided into two categories based on GAN and feature transformation.(2) Model-based methods. Model-based methods are divided into three categories: transfer learning, metric learning and meta-learning. And discuss the advantages and disadvantages of each classification method. Finally, the results are summarized and the future development direction of few-shot network traffic classification technology based on deep learning is prospected. © 2023 SPIE.

关键词： Deep learning

来源：评论

学校读者我要写书评

暂无评论

PFCA: An Intelligent Method of Multidimensional User Profile Construction and Application 7

PFCA: An Intelligent Method of Multidimensional User Profile...

引用

7th International Conference on Big Data Analytics, ICBDA 2022

作者： Xia, Yi Zhou, Gang Luo, Junyong Lan, Mingjing Chen, Xiaohui Wang, Shiyu State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (纸本)9781665479387

User profile plays an important role in analyzing high influencers in the era of big data. For example, the profile can help the political strategists to scrutinize the performance of a candidate or party and overcome their weakness long before election day. In this work, we propose a novel profile construction and application model (PFCA) that aims to discover the overall character of high influencers, and apply PFCA to Donald Trump, one of the most influential people recently, to identify Trump's comprehensive personalities. We first collect and preprocess profile data from static and dynamic aspects of multi-source data, and then categorize multi-source data topics and emotions. Next, we attempt to construct the profile from four aspects: fundamental characteristic, behavior characteristic, value characteristic and situation characteristic, to explore the correlation among these characteristics. Finally, PFCA is verified by the outcomes of the 2016 and 2020 U.S. presidential elections, and is evaluated by relevant experts, which proves that PFCA is relatively effective and efficient to profile high influencers. Moreover, PFCA can be further applied to related analysis and prediction in the 2024 U.S. presidential election. © 2022 IEEE.

关键词： Sentiment analysis

来源：评论

学校读者我要写书评

暂无评论

A Research and Verification of MFC Framework-aware Virus Infection Technology 3

A Research and Verification of MFC Framework-aware Virus Inf...

引用

3rd International Conference on Computer Vision, Image and Deep Learning and International Conference on Computer engineering and Applications, CVIDL and ICCEA 2022

作者： Guo, Wang Li, Yang Sha, Zihan Bu, Wenjuan State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (纸本)9781665459112

Portable Executable (PE) virus technology is an important threat factor in the field of network security. With continuous upgrading of malware detection technology, virus infection technology will also gradually evolve from infecting the structure of the PE file to infecting the deeper structure of the program, and thus achieve higher level of stealthiness and confrontation. This paper proposes a structure-aware virus infection technology for Microsoft Foundation Classes (MFC) software framework, aiming to explore a new method of virus infection. This research aims to verify the virus infection method of MFC program framework perception through experiments, and then provides a solution to the new PE virus, which can help improve the current level of heuristic malware detection technology. © 2022 IEEE.

关键词： Viruses

来源：评论

学校读者我要写书评

暂无评论

A Malware Classification Method based on Attentive Bidirectional Model 7

A Malware Classification Method based on Attentive Bidirecti...

引用

7th International Conference on Intelligent computing and Signal Processing, ICSP 2022

作者： Gui, Hairen Liu, Fudong Zhang, Chunyan Tang, Ke State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (纸本)9781665478571

Deep learning techniques have achieved significant improvements in malware analysis application, such as malware detection, malware classification and vulnerability mining. Convolutional neural network (CNN) is applied to extract local features of samples, while recurrent neural network (RNN) mainly extracts global features from sequence context. However, the pooling layer in CNN is a nonlinear dimensionality reduction operation, which can cause a loss of correlation feature. The raw RNN or LSTM model cannot perform parallel computing when calculating the feature vectors, limiting the effect of model application. Hence, we propose a novel malware classification model named BGSA (Bidirectional GRU with Self- Attention), which applies bi-direction gated recurrent unit (BiGRU) and self-attention mechanism. To demonstrate that the model can improve the feature extraction capability, a series of comparison experiments have been conducted on the VxHeavon dataset. The results show that the classification of BGSA reaches 92.54% in term of precision, which surpass the performance of method based on LSTM+CNN and BiGRU. © 2022 IEEE.

关键词： Long short-term memory

来源：评论

学校读者我要写书评

暂无评论

An Abnormal Domain Name Generation Method Based on a Character-Level Model 4

An Abnormal Domain Name Generation Method Based on a Charact...

引用

4th International Conference on Robotics, Intelligent Control and Artificial Intelligence, RICAI 2022

作者： Wu, Chunbing Fei, Jinlong State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (纸本)9781450398343

Internet technology develops by leaps and bounds, new types of botnet attacks are springing up. Currently, most malware are stealthy when implementing attacks since they can bypass detection strategies such as blacklists and sinkholes. These malware mainly use DGA to generate many random domain names to prevent C&C server IP addresses from being traced to the source. To resist botnet attacks, machine-learning methods have been proposed for detecting abnormal domain names. However, there are still some inadequacies, e.g., detection methods based on manual feature extraction couldn't identify novel variant DGA domains, and existing neural network detection methods have weak generalization ability and lack training data for new variant DGA domain names. To address the above problems, this paper proposes an abnormal domain name data generation technology based on a character-level model. Firstly, combining specific text generation-related ideas improves the character-level language model. Secondly, the Bi-GRU network model and attention mechanism are used to thoroughly learn the internal correlation features, and simulate the generation of abnormal domain names. The experimental results show that this method's abnormal domain name data is highly similar to the actual domain name in relative entropy and character composition. Furthermore, it verifies the feasibility of the generated data as training data set for a new variant DGA domain name, which to a certain extent, helps to improve the generalization ability of the abnormal domain name detection model. © 2022 ACM.

关键词： Botnet

来源：评论

学校读者我要写书评

暂无评论

A Tree-based Structure Mutation Method for Network Protocol Fuzzing 6

A Tree-based Structure Mutation Method for Network Protocol ...

引用

2021 6th International Conference on Information Science, Computer Technology and Transportation, ISCTT 2021

作者： Yang, Rui Jing, Jing Qi, Xuyan Zheng, Hongbing State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (纸本)9783800757282

It is difficult to fuzz the protocol because the format of network protocol is complex and often involves syntax and semantic constraints. Traditional fuzzing often makes a trade-off between the validity of protocol message and mutation space when applied to network protocol. In this paper, we present a tree-based structure mutation method for network protocols. Firstly, the test template is parsed into a tree structure, then the tree nodes are mutated specifically, and finally the test message is reassembled for fuzzing. To demonstrate the effectiveness of our method, we evaluated it on three protocol programs, and the results show compared with the traditional mutation-based and grammar-based method, the tree-based structure mutation method can further expand the mutation space while ensuring the efficiency of the message, and has a higher probability of discovering program paths and triggering vulnerabilities. © VDE VERLAG GMBH · Berlin · Offenbach.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

A PV-DM-based feature fusion method for binary malware clustering 4

A PV-DM-based feature fusion method for binary malware clust...

引用

4th International Conference on Communications, Information System and Computer engineering, CISCE 2022

作者： Gui, Hairen Zhang, Chunyan Huang, Yizhao Liu, Fudong State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (数字)9781665498487

ISBN: (纸本)9781665498487

Utilizing NLP methods in malware classification or clustering applications is a hotspot in malware analysis. Binary malware embedding has become an important direction and basis for malware analysis. Existing binary malware vectorizing methods cannot extract semantic and functional information simultaneously, which may affect the performance of family clustering. In this paper, a novel feature fusion representation method is proposed based on PV-DM (Paragraph Vector-Distributed Memory) model, and the K-Means algorithm is adopted for family clustering. The method combines the features of program-level API call sequence and function-level instruction sequences by means of PV-DM. The functional and semantic representation capabilities of binary malwares can be improved. Malware public dataset VXHeavon is taken as the dataset. A series of experiments are conducted, using the existing family labels as the clustering accuracy evaluation standard. The results demonstrate that our feature fusion representation method outperforms the baseline method by 4.96% in terms of accuracy. © 2022 IEEE.

关键词： Malware

来源：评论

学校读者我要写书评

暂无评论

A Survey of Available Information Recovery of Binary Programs Based on Machine Learning 5

A Survey of Available Information Recovery of Binary Program...

引用

5th IEEE International Conference on Artificial Intelligence and Big Data, ICAIBD 2022

作者： Shao, Wenqiang Yang, Qichao Guo, Xixi Cai, Ruijie State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (纸本)9781665499132

A binary program is an executable file obtained by compiling the source code. In order to reduce the size of software or prevent reverse engineering, software publishers usually only release executable files that have stripped a lot of source information. The rich source-level information is very helpful for software hardening and malware analysis. In recent years, the rapid development of artificial intelligence has promoted the research of binary program information recovery. This paper summarizes the relevant research on effective information recovery of binary programs since 2016. Firstly, the importance of recovering source-level information from binary programs is introduced. Next, the problems are divided into three categories, compiler traceability, signature information recovery and variable information recovery of source program. Finally, based on the latest technology in this field, the future development direction is analyzed. © 2022 IEEE.

关键词： Machine learning

来源：评论

学校读者我要写书评

暂无评论

Multi-protocol comprehensive topology probe technology under non-cooperative conditions

Multi-protocol comprehensive topology probe technology under...

引用

2022 International Symposium on Computer Applications and Information Systems, ISCAIS 2022

作者： Lu, Xuanting Yang, Qichao Liu, Shengli Yang, Pengfei State Key Laboratory of Mathematical Engineering and Advanced Computing Zhengzhou China

ISBN: (纸本)9781510655010

With the rapid development of the Internet and the increasingly complex and diverse network environment, it is of great significance to quickly and accurately obtain the topology environment of the target network for optimizing network management, locating network resources and ensuring network security. Most of the existing network topology detection methods are implemented based on the SNMP protocol, or use the Traceroute-like method for active detection. However, in a non-cooperative environment, that is, the router disables SNMP permissions or prohibits responding to probe packets to shield Traceroute probes, network probes will face problems such as missing information and blocked probes. Based on the original SNMP protocol detection, this paper adopts RIP, OSPF and other routing protocols to supplement the detection method, which greatly reduces the number of undetectable nodes. In addition, for a small number of router devices that are still undetectable, the Traceroute directional breakthrough method using the routing table as a clue is used to bypass these undetectable nodes, complete the missing information in the network topology, and further optimize and improve the detection results. © 2022 SPIE

关键词： Network topology

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：